XA9952646

SMiRTl3 IAEA-WS August 1995 Iguasu, Argentin

On Southern Hyogo-prefecture Earthquake and Some Related Activities in

Heki SHIBATA, Professor-Dr.

Mechanical Eng'g. and Materials Sci. Faculty of Engineering Yokohama National University 156 Tokiwadai, Hodogaya, Yokohama 240, JAPAN Fax 81/45-331-6593

Abstract

This paper consists of three parts. At first the reporter discusses on the earthquake event on January 17, 1995, and then on the summary of the report of examining the adequecy of the guideline of seismic design of nuclear power plants in Japan by the task group, Nuclear Safety Commission. And also on the activity of "the sub-committee on the research of seismic safety" for the future research subjects during 1996 ~ 2000 F.Y.

Part 1 On the Southern Hyogo-prefecture Earthquake

1.1 Introduction The event occured at 5:46 am on January 17, 1995 was one of the most serious earthquake disasters since the Kwanto earthquake in 1923 in Japan. In the Kwanto earthquake, approximately 140,000 were killed in and Yokohama area. Most of casualties were caused by the extended fires in Cities of Tokyo and Yokohama. The direct deaths are estimated as 10,000. In the earthquake disaster of this time, officially called as "Hanshin-Awaji Great Earthquake Disaster", approximately 5,500 were killed, and those were directly caused by structural failure mostly. One of the serious natural disasters, which caused such a big number of deaths is "Isewan Typhoon and High-tide" in 1959.

37 As seismic events, this earthquake is one of the greatest ones, may be once or every century through Japan. In this area, we experienced several destructive seismic events since 8 Q. but they might be not so serious ones according to historical records including 16 9 one: which is famous. Some one estimated its return period might be more than 5,000 years (in Fig.8). The characteristics of its ground motion was very unique, that is, very high velocity as well as high acceleration and short duration (Table 1). This caused many casualties and serious, very uneque structural failures. The feature of the ground motion, damage of various kind of structures, and other unique events will be described in the following chapters, which were presented at SMiRT 13 in Porto Alegre last August. Of course, this area is one of the most densely populated and highly industrialized areas. The western part of Kei-han-shin area, which is the second busiest area in Japan, is centered by City of . This is the reason why this event is called as Kobe earthquake. Officialy. Southern Hyogo-prefecture earthquake 1995 and the disaster itself is called as Hanshin-Awaji Great Earthquake Disaster.

1.2 Purpose of Field Survey What are the purposes of the survey on seismic damages? How do we learn the lessons from the seismic event. Usually we start immeadately after the event from sur- veying what are happening and how they are. For this, video is a very powerful tool now. The author started its recording a half hour later of the event when he recognized that the event had been very large scale event, and had been continuing for six hours. The major items and purposes are as follows : 1) How are there any phenomenon or failure mode which has been never observed:' In this case, they will be reported as examples in the field. 2) How adequate the models, which have been used for the design, are proven or verified through their behaviors observed? 3) How the statistics of structures and other systems are, for example, damage ratio of underground pipeline etc.? 4) How to recover structures and systems from their damaged state in a short range, hours and day, and in long a range, month and year? 5) How did or does the event give the impact to individuals and the society?

1.3 Facts on the Event Magnitude : M = 7.2 Focal Depth : H = 14 km

38 Time and Date : 05:46 , 1/17/1995 , JST Casualties : 5,500 deaths + 35,000 injures Damaged Buildings & Houses : App. 160,000 . Estimated Total Loss : $ 100 Billion Those numbers on damages are still moving

1.4 Fault, Ground Motion and Intensity The slips of related fault have been studied by seismologists, and various kind of new- facts have been being found. One of the features is high velocity ground motion. Japanese Meteological Intensity Seale exceeded VI in this event. This lias been the first event since it was denned after Fukui earthquake in 1948. The definition of Intensity VII is as follows : / = VII > 400 gal and Damage rate of buildings > 30 % There were several issues, for example, how to defined the damage rate and so on this time, but JMA decided that in the area of the center of Kobe City the intensity was VII. Following issues on their feature of ground motions which were made immeadiatly after the event, now some of them are clear, for example : How strong the effective PGA? Is the recorded maximum ground velocity reliable? Which faults did cause main shocks? The short duration of ground motion, is very unique as Japanese eartuquake. And initial several peaks were significant to cause structural damages as shown in Fig.l. and Table 1. Main shock consits of three shocks, and the waves were focused into the eastern part of the City Fig.2. Distribution of peak ground accelerations is shown in Fig.3. And their attenation curves are shown in Figs.4 and 5. Through the activities of the seismologists after the event, we feel the vecessity of the establishment of the engineering seismology for estimatimation of local ground motions, that is, micro-zoning.

1.5 Damage State of Various Kind of Structures i) Structures Highway Bridge : collapsing and sliding Building ; High-rise Building : less damage

39 Reinforced Conventional Building : large damages, three modes ; lower level collapse, one particular level collapse, and PS effect failure Steel Building : large damages, especially degraded one and brittle failure Wooden house : collapsing (and burnt) Piles ; failure by shear force mainly Pier and Embankment : side slip and subsidence Tower : less resonance-type failure in the area Tank : liquefaction effect, and sloshing in outside of Kobe area Comuter : overturn and cable failure Lifelines : various mode ii) Mechanism Single shock failure Resonance failure P-<5 effect Liquefaction Brittle Failure Mechanism of brittle failure has not been known well. However, the understanding on the phenomenon is diverged at each field as shown in Table 2.

1.6 Time-history and Damage The time history, that is, the partterns of ground motion of this event are very unique, and it has been proved by seismologists that there are exact reasons to induce such ground motions. The features of this event are quite different from other destructive earthquakes which have been recorded in Japan since 1880's instrumentally. The most serious destructive earthquakes, which were recorded in the past, are inter- plate type huge earthquakes and their epi-centers were in the ocean and their epicenter distances are usually more than 100 km. Durations of such earthquakes were over one minute in general. They induced resonance type failure to structures and this S-phase was followed by the surface wave period. Main points are as follows : i) Very short duration from initiation of ground motion to main peaks ; approximately 2 sec as shown in Fig.l and Table 1. ii) Similar wave form of acceleration to displacement :

40 This means, that the waves consist of rather simple component distribution, and it makes easy to analyse them from seismological view-points. Comparison of NS motions at Kobe Ocean Observatory to the images of Video, recorded at 10 sec ahead at the NHK office in the area shows the process of events as Table 1. The shaking tests of human on the table, demonstrate the strong effect to human body as well as to structures. The large amplitude of displacemet in a short duration may cause many unique features of this event, especially, many deaths. It should be mentioned that such a type of ground motions may be expected more in low seismicity area rather than high seismicity area, even though the probability of occurence might be low.

1.7 Response Spectrum and Particle Motions. Role of Vertical Component They have been discussed, but their direct effect has not known exactly yet. Some features are as follows : i) Dominant in longer period range, 1 ~ 1.5 sec in the response spectrum (in Fig.9 and 10) ii) Corelation of horizontal motions and vertical motions is recognized iii) Video-recordings at super-markets have a certain role for seismological study iv) Behavior of box-shape articles, over-turning, sliding and jamping As far as the second item, we feel the strong necessity of more study.

1.8 Damage of Industrial Facilities There were very serious damages of industrial facilities such as ship-builders, macline factories, habor facilities, plants, lifeline and so on. Most of them came from liquefaction rather than acceleration effect. Therefore, those were found more in areas near to the coast line compared to ordinary buildings and residential houses. 1) Effect of liquefaction and land-sliding : i) Overhead piping system above-ground deformation ii) Crane failure including largesize container crane iii) Settlement of heavy articles iv) Deformation of rails for O.H. crane and other traveling machines v) Ship-builders' yard and dock vi) Settlement of cylindrical tank vii) Lifeline, underground pipings viii) Switching station

41 2) Accerelation effect i) Railway train derailing ii) Power plant, boiler and pipings iii) Switching station and transmission line iv) Towers-type crane v) Tower for micro-wave transmission and other purpose

1.9 Damage Modes. Newly Observed The following modes were observed. Some of them, especially items ii) and v) are not known on their exact reasons. i) Land-slide in flat ground and coastal area ii) Brittle failure of steel column iii) One-direction shear failure of R.C. column iv) New-type cracking of brittle material v) Mid-story collapsing of building — propagation wave model vi) Partial failure of structure, and electric equipment and distribution system caused by other failed structural member or element vii) Overturning of buildings by P-£ effect, especially R.C. buildings viii) In some places vertical acceleration are higher than horizontal one, and both are corelated ix) Low acceleration near to surface fault and very high acceleration in regions where slighty for from the fault where surface fault is not found. Those items shall be studied intensively, and also their damage survey systems must be developed.

1.10 Standardization of Design Basis Earthquake Central Commission for Disaster Prevention, Chairman : Prim-minister, issued "the new fundamental principle for disaster prevention" on July 18, 1995 based on the facts which we experienced. In this article, two levels of counter-measures, including Design Basis Earthquake are stated. 1) For any events which are expected to occure several times through its life, it must be remaining as functional without significant damage. 2) For any rare event, human lives must be kept without serious damage. This comes from the original principle of our nuclear power plant design. There are approximately 40 seismic codes in Japan. These codes must be modified and unified

42 according to their fundamental concept and philosophy. Following items will be reviewed next two years. i) Factor of Importance ii) Concept of Zoning iii) Final Protection for the Safety iv) Standard Design Response Spectrum v) Vertical Ground Motion for the Design vi) Longer-period-range Ground Motion vii) Duration of Ground Motion viii) Measure for Rare Catastophic Event and its Level ix) Active / Capable Fault Protective Practice x) Liquefaction Protection xi) Land-slide Protection in Flat Area near Coastal Line xii) Land-slide Protection for Large Scale Slope xiii) Allowable Damage and Loss in Various Cases xiv) Insurance and Financial Back up by Official Organization xv) Mitigation of Social Impact

1.11 Liquefaction and Sliding Most of damages of industrial facilities were induced by liquefacion and sliding of embankment and pier. Soil, mixed with gravel and silt, caused liquefaction phenomenon by very high response of soil layer, and sliding ground toward the sea. The phenomenon of liquefaction was said to occure with only uniform granular sand.

1.12 Conclusion and Recommendation 1) This event, Kobe earthquake, is a really rare event. 2) There are very unique and unexpected features in the view points of seismological studies. 3) It is very difficult to predict its features. 4) Safety design must be done in consideration with two levels of events as described in the previous chapter. 5) The level of the severer event must be settled based on the knowledge on rare events. However, it is very difficult to develop it only by engineers, and they must ask the positive assistance of scientists.

43 6) The design for safety systems under seismic conditions is very significant, and it must be carefully examined in the sens^ of system dynamics. 7) P-6 effect of steel frame and ductile R.C. buildings were observed, and they are some different from ordinary ones. 8) The cooperation between seismologists and engineers including safety engineers must be encolleged. a) Continuous efforts to improve models for design must be done based on facts which we obser%'ed.

Part 2 The-State-of-the-Arts of Seismic Design of Nuclear Power Plants in Japan.

2.1 Introduction After the event, the Nuclear Safety Commission organized a task group, chaired by Professor Kojima. This TG consisted of nine specialists including the reporter. They examined the following points to clarify that the Guideline for (examining) the Seismic Design of Nuclear Power Plants in 1981 NSC, which has been used for the regulatory purpose, is adequate: i) If a nuclear power plant is constructed according to the Guideline in Kobe Area, the design would be reasonable? ii) How S2 ground motions would be strong? iii) How the response spectrum of S2 would be conservative? Nine meetings and one field survey were made.

2.2 Design Basis Earthquakes in Japan Two levels of "Design Basis Earthquakes" have been employed in Japan since almost beginnings, 1960's. In the Guideline, Si and S2 are defined as follow : Si, the strongest earthquakes, and S2, the limit earthquakes. The second one may be interpreted as the upper-bound earthquake. Si earthquake may be the historical maximum earthquake in the site region. The historical records on destructive earthquakes have been kept since 5C, and those since 9 ~ 10$ can be listed in the seismic catalogue. The number of them from 416 to 1882, non-instrumental era listed is 267 and from 1884 to 1993, instrumental era, is 163. Some compensations to decide Si earthquake are made based on seismological knowledge, because, their return periods in some regions may be more than 1000 years, for which we can find historical written recods. And the periodical change of frequency of occurence might be observed in some area.

44 So earthquake is the upper-bound earthquake, whose magnitude M can be estimated in region by region as shown in Fig.6. According to the practice in Japan, the annual probability of occurence of a certain earthquake doesn't follow the stochastic relation in the stronger level, and there is the limitation as explained in Fig.7, and this value can be estimated by the seismotechtonic structure of the region (in Fig.6). However, level of intensity may be more diverged.

2.3 Comparison of Design Basis Earthquakes to the Kobe Earthquake According to the requirement of the Guideline. S2 should be as follows : M = 7| A = 7 km If we assume that a nuclear power plant in the area M95 = 7.2 and A95 = 16 km based on Japan Meteorogical Agency. Those values, officially reported, are formalized as the definition as to concentrate to one focus. The survey result by the seismologist the situation is more complicated. The focal distance Rg^ can be defined not so clearly. The above value was decided under consideration of such a situation, and the value based on original definition, it might be more than 50 km. A is the epicenter distance, and the focal distance is the distance to the focus of the event. However, the definition of a focus is the point which the initial slip of a fault movement had started. In the case of this event, its depth H is estimated as 14 km. S2 earthquake in the Guideline is defined in two ways : the maximum earthquake which might occure at the point estimated by seismological survey on active fault distribution, and that ; M = 6.5 underneath of the site, that is A = 7 km, H = 7 km. In this case, the former definction is applicable.

2.4 Response Spectrum of the Event Even though there are many records of strong seismograph in the area, those of rock site are limited. The site of a nuclear power plant in Japan must be rock site. Therefore, the records observed in the tunnel at the campus of were only met to this requirement. These the design basis spectra, which are called as Ohsaki spectra based on approxi- mately 40 ground motion records in rock site observed in the world wide, are the standard spectrum which is recommended in the siipplimental explanation of the Guideline. Ac- cording to the response spectrum, in the lower frequency region, the response spectra of them are dominated compare to the design basis response spectra wliich are used for the design (in Figs 9 and 10). The comparison of this standard spectra to the response spectra of ground motions in the area was made, and the standard spectra are more conservative than the spectra of the event at the Kobe University except in the lower fre- quency region than 06 sec. Eigen-periods of significant structures, piping systems and

45 equipment are generally shorter than this value. This means that the standard spectra for the design are adequate in the view point of the margin.

2.5 Vertical Ground Motion It seems to be that many evidences of domination of vertical ground motions of this event. One of the reason comes from nonlinear characteristics of soil layer for transmission of S-wave, and linear for that of P-wave. The ratio of peak value of vertical ground acceleration to that of horizontal one was less than a half, 1/2, in general. Also, that of peak velocity and spectral intensity are the same situation. This means that the ratio used for the design is also adequate. Even though, we could find many facts of behaviors showing the vertical ground motion were dominated as mentioned above, and it might be induced by their corelation. Study on the corelation of both vertical and horizontal motions shall be necessary.

2.6 Remarks The analyses of key structures and items in a typical power plant based on the spectra had been made as a reference, and we could not find any critical issue. The report on the new basic proposal prepared by the Central Commission of Disaster Prevention pointed out that all kind of structures, not only nuclear facilities, must be designed in two levels, such as the concept of Sj and $2- The draft of the sped?! committee, the Science Council of Japan recommended that the equalization of the concept of seismic codes, whose number is approximately 40 kinds in Japan, must be made. In 1981, only the concept of zoning had been established except for nuclear power plants, however, now there are opinions that the activity of local active or capable fault must be considered in the codes of others. Also the continuous effort to improve the seismo-techtonic map like Omote map is necessary.

Part 3. Future Researches on Seismic Design of Nuclear Power Plants in Japan

As a part of the safety study program, the subjects and programs of seismic safety studies by various organizations will be reviewed every five years. For this, a sub-committee was organized under the chairmanship of Shibata, the reporter, and this belongs to the Committee on Nuclear Safety Research, NSC. In this year, next five year projects, 1996 ~ 2000 F.Y. have been reviewed. This job was started in October 1994, therefore, on the way of process the Southern Hyogo-prefecture earthquake occured. As a result, the sub-committee completed their Teport under the condition that the report will be reexamined in next one year, because there might be various topics arising based on facts which were observed in the event. Some of them are as follows :

46 i) How is the distribution of the maximum velocity of ground motions surrounding the fault line? ii) How is the map of the upperbound magnitude through Japan? iii) How is the peak ground acceration of vertical motions affected by local movement of the fault? iv) How is the structural response of buildings, piping systems and equipment to such vertical ground motions, and how is related to the horizontal motions? v) How is the brittle failure of ordinary steel using for piping support and others7 Those subjects would be discussed in future meetings with other subjects newly arising according to the impression of the reporter. And they will be discussed in other fields also. And more over, it might be necessary how the level of the serious but rare event shall be assume for the safety study.

Acknoledgement The manuscript for the Proceedings of Post Conference Seminar 16 of SMiRT 13 which was held on August 21 and 22 at Iguazu Argentina was required to he conpleted by the end of November. Therefore, the author trys to rewrite it with new informations and his consideration based on his paper distributed at the CSNI meeting, OECD which was held on November 29 and 30, 1995 in Paris. And also this material was distributed at the IAEA-INS Workshop in Taejon, Korea, on December 6 and 7, 1995. The author greatly appreciate the advices of Dr. Tcbioha, JAERI and Mr. Miller, 9ECD: NEA secretriat, and related Task Groups.

47 Digital Memory Device for 10 second-ahead recording Video i = small articles (books) on "he desk are slightly moving.

a = large motions of articles are observed.

b = overturn of file-cabinet.

c = TVset is droning from a rack.

900 «~:N00C£.

AC1. r-MflW

.900-

Table 1 Sequence ofFailures Observed in Kobe Branch, Nippon Broad Casting Corp. by Video.

Table 2 Mechanism of Understanding of Brittle Failures.

* Phenomenon is understood for 1) Nuclear P.V. Specialist: It had been known in early '80s. 2) Ship Builders' Specialist: It has been steel doubtful on its mechanism. Some test results are known, and Tech. Comm. in WES. J. has been organized last two years. 3) Structural Engineering Specialist: It is now as a Subject Newly faced on.

* Fact NDT Temperature shifts by plastic deformation in some case. Ductility factor, fi = 1.20, may push up the NDTT to Room Temperature in some cases. Temperature of the morning in Kobe, on Jan. 17, 1995, is estimated at 0 ~ 4°C

source : Prof. H. Kobayaslii Tokyo Inst. of Tech. 48 T,..i;: :.«^i;;

Fig.l Time History observed at Kobe Ocean Meteorogical Observatory.

Hyogo 1995/01/17

t-\a - 2.5 i iO"25 dync-cr : 5.9 Dtp:

Fig.2. Mechanisms of sequence of Main Shocks from Far-distance Record Analysis (AnaJized by Prof. Kikuchi, Yokohama City Univ.).

49 Unit gal

Fi°;-3 Distrubution of P.G.A.

50 ; KSf.WCMIVA >

-

1 me . ! SHIN-

LJ t-J 30. C

"•I

i ! i i i J 1Q.C

i_

I : : . . , I i--.-s. .1 C.2 0.5 1.0 2.0 5.0 10.0 20.0 SCO 1C0.C 200.0 500.0

DISTANCE fXM3

Fig.4 Attenuation Curve of P.G.A.

f,'-:=7.0

2:

0.2 0.5 !.O 2.9 5.0 !0.0 20.0 50.0 1C0.0 200.0 5C0.0

OISTP.vCE (KM)

Fi°;.5 Attenuation Curve of P.G.V. 51 Fig.6 Map of Distribution of Maximum Potential Earthquake in Japan

DBE Concepl in Japan

t-Historical Max. { o* ^ Now)

S, S2 1 {CJ Design Basis Earthquake

Fig.7 Relation of Si and S2 Design Basis Earthquakes To Probability of Occurrence of Seismic Event

52 9S- 9 iooa

99- 6 - 500

99- 5 1 200

- !00 ar. UJ

O UJ UJ

S 10 20 SO 100 200 ACCELERATION ( CflL )

0.996? NON-EXCEEOBNCe 58.

500 I 1 1 1 • J i ' i-^ I i ' ' it 1 I Aoo\ — f— — — I 0 • I 1 0- 95 ~ I I -. 1 ^^ 1 200 CL 9

>"- 0- 8 I 1 ^"^ ' r~ -\ g _ I 1 ^s _ I -

jS^ ->^ ^^ 1 1 UJ

i >

20 i i_ i_ i i •^%^ i i i t >^ 1 i i 10 l_ I 1 1 - j 1 1 1 - 1 I 1 1 - - i I 1 1 - III!. 1 1 1 I 1 ! 1 , 1 < 10 20 50 100 200 500 r vcoo i

Fig. 8 Return Period of Earthquake vs. Estimated Acceleration in Kobe area

53 FREQUENCY 0

(h-0. 05> 2000

1000 -

8

0.01 0.1

PERIIJO (SEC)

0.1 Fig. 10 Comparison of Response Spectrum, Kobe v.s DBE (Scimi-log)

0. 02 PERIOO (SEC)

Fig. 9 Comparison of Response Spectrum, Kobe vs DBE (log-scale)